renumbering (was: 6BONE AUP)

[bmanning@ISI.EDU]

> > We actualy went through a global renumbering session on the 6bone once,
> > when we moved from 5F RFC1897 addresses to the new aggregated 3FFE ones.
> > 
> > Transiting my site was not that big a deal. The main issue was to convince
> > people running 6bone leaf sites to do the job.
> > The overall process took more than 3 months at the time.
> > 
> > In the RIPE draft paper about IPv6 address allocation, they specify
> > that you will have 3 months to renumber.
> 
> In practice, if you have three months, the tools themselves have to
> make it possible for someone to accomplish the job in a couple of days 
> because they won't, in practice, do much until the last possible
> minute.
> 
> Renumbering semi-frequently on the 6bone would probably encourage us
> to make the tools better and to stop thinking of IP addresses as
> stable and permanent. (Look at the world's ntp.conf files if you want
> to see where that leads...)
> 
> Perry

Much of the thinking wrt renumbering went into the abortive PIER wg.
NTP, DNS, SNMP(network mgmt) are all problem areas.  Pop up a level
and things like NFS mount points, Web Caches and SDR/Confctl tools
are also suspect.

I'll point out that the first pass at renumbering the 6bone did not,
for the most part, involve DNS servers. Most of them were (and still are)
only IPv4 aware.

-- 
--bill